Airbnb multicity

Airbnb Dataset¶

This dataset consists of approximately 3.1 million Airbnb listings collected between June 2022 and May 2023 across 80 cities worldwide. It includes geographic location, property characteristics, host activity, and review metrics. For the benchmark, a cleaned subset from six cities—Paris, Rome, London, Amsterdam, Melbourne, and New York City—was selected.

Run this notebook in Google Colab:

Remember to install the srai library before running the notebook:

%pip install srai[all]

In [1]:

Copied!





# plotting imports
import contextily as cx
import seaborn as sns
from matplotlib import pyplot as plt
from matplotlib.lines import Line2D
from matplotlib.patches import Patch

# dataset import
from srai.datasets import AirbnbMulticityDataset
# plotting imports
import contextily as cx
import seaborn as sns
from matplotlib import pyplot as plt
from matplotlib.lines import Line2D
from matplotlib.patches import Patch

# dataset import
from srai.datasets import AirbnbMulticityDataset

In [2]:

Copied!

airbnb_multicity = AirbnbMulticityDataset()
airbnb_multicity = AirbnbMulticityDataset()

In [3]:

Copied!

type(airbnb_multicity.train_gdf), type(airbnb_multicity.test_gdf)
type(airbnb_multicity.train_gdf), type(airbnb_multicity.test_gdf)

Out[3]:

(NoneType, NoneType)

Loading default version

In [4]:

Copied!

ds = airbnb_multicity.load()
ds.keys()
ds = airbnb_multicity.load()
ds.keys()

Out[4]:

dict_keys(['train', 'test'])

In [5]:

Copied!

type(airbnb_multicity.train_gdf), type(airbnb_multicity.test_gdf)
type(airbnb_multicity.train_gdf), type(airbnb_multicity.test_gdf)

Out[5]:

(geopandas.geodataframe.GeoDataFrame, geopandas.geodataframe.GeoDataFrame)

In [6]:

Copied!

print("Aggregation H3 resolution:", airbnb_multicity.resolution)
print("Aggregation H3 resolution:", airbnb_multicity.resolution)

Aggregation H3 resolution: 8

In [7]:

Copied!

print("Prediction target:", airbnb_multicity.target)
print("Prediction target:", airbnb_multicity.target)

Prediction target: price

In [8]:

Copied!

gdf_train, gdf_test = ds["train"], ds["test"]
gdf_train, gdf_test = ds["train"], ds["test"]

In [9]:

Copied!

print("Available cities:", sorted(gdf_train["city"].unique()))
print("Available cities:", sorted(gdf_train["city"].unique()))

Available cities: ['amsterdam', 'london', 'melbourne', 'new-york-city', 'paris', 'rome']

In [10]:

Copied!

gdf_train.head()
gdf_train.head()

Out[10]:

	id	name	host_id	host_name	neighbourhood	room_type	price	minimum_nights	number_of_reviews	last_review	reviews_per_month	calculated_host_listings_count	availability_365	number_of_reviews_ltm	city	date	geometry
0	9396150	Cosy studio sur cour fleurie. Métro Saint-Man...	5783938	Alexandra	Ménilmontant	Entire home/apt	53.0	7	7	2022-06-02	0.34	1	308	5	paris	2022-06-06	POINT (2.42054 48.84782)
1	638764995602497687	JOLI T2 LUMINEUX de 34 m² à 7 min de PARIS	460853253	Sawsen	Buttes-Montmartre	Entire home/apt	59.0	1	0	None	NaN	1	13	0	paris	2022-06-06	POINT (2.3591 48.90649)
2	23135649	Charming studio, Pont de Neuilly - Paris	171938056	Julien	Passy	Private room	65.0	3	31	2022-05-31	0.65	1	4	6	paris	2022-06-06	POINT (2.25285 48.88173)
3	54220288	Superbe appartement moderne - Les Docks Saint-...	141681595	Théo	Batignolles-Monceau	Entire home/apt	103.0	3	4	2022-05-22	1.08	1	214	4	paris	2022-06-06	POINT (2.32833 48.90859)
4	885179	Parisian luxury apartment ...	4176034	Bernard	Buttes-Montmartre	Entire home/apt	65.0	3	84	2022-03-19	0.99	1	245	21	paris	2022-06-06	POINT (2.33163 48.90531)

In [11]:

Copied!





fig, axes = plt.subplots(
    2, 2, sharex=False, sharey=False, figsize=(12, 15), width_ratios=[3, 1]
)

cities = [("Amsterdam", 0.05), ("London", 0.01)]
for row_idx, (city_name, marker_size) in enumerate(cities):
    city_train = gdf_train[gdf_train["city"] == city_name.lower()]
    city_test = gdf_test[gdf_test["city"] == city_name.lower()]
    train_points = len(city_train)
    test_points = len(city_test)
    train_pct = 100 * train_points / (train_points + test_points)
    test_pct = 100 * test_points / (train_points + test_points)

    ax_map = axes[row_idx][0]
    city_train.plot(color="orange", markersize=marker_size, ax=ax_map, label="train")
    city_test.plot(color="royalblue", markersize=marker_size, ax=ax_map, label="test")
    ax_map.set_title(
        f"{city_name} data - points on a map"
        f" (Train: {train_points} ({train_pct:.2f}%),"
        f" Test: {test_points} ({test_pct:.2f}%))"
    )
    ax_map.legend(
        handles=[
            Line2D([], [], marker="o", color="orange", linestyle="None"),
            Line2D([], [], marker="o", color="royalblue", linestyle="None"),
        ],
        labels=["Train", "Test"],
    )
    cx.add_basemap(
        ax_map, source=cx.providers.CartoDB.PositronNoLabels, crs=4326, zoom=12
    )
    ax_map.set_axis_off()

    ax_dist = axes[row_idx][1]
    sns.kdeplot(
        x=city_train[airbnb_multicity.target],
        label="train",
        color="orange",
        ax=ax_dist,
        fill=False,
        cut=0,
    )
    sns.kdeplot(
        x=city_test[airbnb_multicity.target],
        label="test",
        color="royalblue",
        ax=ax_dist,
        fill=False,
        cut=0,
    )
    ax_dist.set_title(f"{city_name} data - target distribution")
    ax_dist.legend()

plt.tight_layout()
plt.show()
fig, axes = plt.subplots(
    2, 2, sharex=False, sharey=False, figsize=(12, 15), width_ratios=[3, 1]
)

cities = [("Amsterdam", 0.05), ("London", 0.01)]
for row_idx, (city_name, marker_size) in enumerate(cities):
    city_train = gdf_train[gdf_train["city"] == city_name.lower()]
    city_test = gdf_test[gdf_test["city"] == city_name.lower()]
    train_points = len(city_train)
    test_points = len(city_test)
    train_pct = 100 * train_points / (train_points + test_points)
    test_pct = 100 * test_points / (train_points + test_points)

    ax_map = axes[row_idx][0]
    city_train.plot(color="orange", markersize=marker_size, ax=ax_map, label="train")
    city_test.plot(color="royalblue", markersize=marker_size, ax=ax_map, label="test")
    ax_map.set_title(
        f"{city_name} data - points on a map"
        f" (Train: {train_points} ({train_pct:.2f}%),"
        f" Test: {test_points} ({test_pct:.2f}%))"
    )
    ax_map.legend(
        handles=[
            Line2D([], [], marker="o", color="orange", linestyle="None"),
            Line2D([], [], marker="o", color="royalblue", linestyle="None"),
        ],
        labels=["Train", "Test"],
    )
    cx.add_basemap(
        ax_map, source=cx.providers.CartoDB.PositronNoLabels, crs=4326, zoom=12
    )
    ax_map.set_axis_off()

    ax_dist = axes[row_idx][1]
    sns.kdeplot(
        x=city_train[airbnb_multicity.target],
        label="train",
        color="orange",
        ax=ax_dist,
        fill=False,
        cut=0,
    )
    sns.kdeplot(
        x=city_test[airbnb_multicity.target],
        label="test",
        color="royalblue",
        ax=ax_dist,
        fill=False,
        cut=0,
    )
    ax_dist.set_title(f"{city_name} data - target distribution")
    ax_dist.legend()

plt.tight_layout()
plt.show()

$No description has been provided for this image$

Getting aggregated hexagon values

In [12]:

Copied!

train_h3, _, test_h3 = airbnb_multicity.get_h3_with_labels()
train_h3, _, test_h3 = airbnb_multicity.get_h3_with_labels()

In [13]:

Copied!

train_h3.head()
train_h3.head()

Out[13]:

	geometry	price
region_id
882a100dd9fffff	POLYGON ((-73.90034 40.69846, -73.90671 40.697...	127.746667
88195dace5fffff	POLYGON ((-0.42167 51.54068, -0.42839 51.53947...	79.500000
881fb474a1fffff	POLYGON ((2.33465 48.91954, 2.32833 48.91839, ...	112.678571
882a100a8bfffff	POLYGON ((-73.90344 40.83853, -73.90983 40.837...	106.071429
88be6236d1fffff	POLYGON ((145.44505 -37.93223, 145.44622 -37.9...	225.230769

In [14]:

Copied!

test_h3.head()
test_h3.head()

Out[14]:

	geometry	price
region_id
88195da681fffff	POLYGON ((-0.10646 51.57144, -0.11317 51.57025...	121.156250
881969504dfffff	POLYGON ((4.87764 52.33471, 4.87606 52.33041, ...	183.818182
881e80cd93fffff	POLYGON ((12.60114 41.96974, 12.59926 41.96496...	200.000000
88be622e41fffff	POLYGON ((145.49699 -37.65429, 145.49816 -37.6...	199.700000
88be63c9c7fffff	POLYGON ((145.28791 -37.83716, 145.28909 -37.8...	35.000000

In [15]:

Copied!





aggregated_train_data = train_h3.cx[-1.04:0.65, 51.09:51.84]
aggregated_test_data = test_h3.cx[-1.04:0.65, 51.09:51.84]

with plt.rc_context({"hatch.linewidth": 0.4}):
    ax = aggregated_train_data.plot(
        airbnb_multicity.target,
        cmap="spring_r",
        legend=True,
        legend_kwds=dict(
            location="right", shrink=0.9, pad=0.02, label=airbnb_multicity.target
        ),
        figsize=(15, 9),
        alpha=0.5,
    )

    ax.set_axis_off()

    aggregated_test_data.plot(
        airbnb_multicity.target, cmap="spring_r", alpha=0.5, ax=ax
    )

    aggregated_test_data.plot(
        ax=ax, linewidth=0.4, color=(0, 0, 0, 0), edgecolor=(0, 0, 0, 0.4), hatch="+++"
    )

    ax.set_title("London data aggregated to H3 cells")
    ax.legend(
        handles=[
            Patch(edgecolor=(0, 0, 0, 0.8), linewidth=0.1, facecolor=(0, 0, 0, 0)),
            Patch(
                edgecolor=(0, 0, 0, 0.8),
                linewidth=0.1,
                facecolor=(0, 0, 0, 0),
                hatch="+++",
            ),
        ],
        labels=["Train", "Test"],
        loc=2,
    )

    cx.add_basemap(ax, source=cx.providers.CartoDB.PositronNoLabels, crs=4326, zoom=11)
    ax.set_axis_off()

    plt.show()
aggregated_train_data = train_h3.cx[-1.04:0.65, 51.09:51.84]
aggregated_test_data = test_h3.cx[-1.04:0.65, 51.09:51.84]

with plt.rc_context({"hatch.linewidth": 0.4}):
    ax = aggregated_train_data.plot(
        airbnb_multicity.target,
        cmap="spring_r",
        legend=True,
        legend_kwds=dict(
            location="right", shrink=0.9, pad=0.02, label=airbnb_multicity.target
        ),
        figsize=(15, 9),
        alpha=0.5,
    )

    ax.set_axis_off()

    aggregated_test_data.plot(
        airbnb_multicity.target, cmap="spring_r", alpha=0.5, ax=ax
    )

    aggregated_test_data.plot(
        ax=ax, linewidth=0.4, color=(0, 0, 0, 0), edgecolor=(0, 0, 0, 0.4), hatch="+++"
    )

    ax.set_title("London data aggregated to H3 cells")
    ax.legend(
        handles=[
            Patch(edgecolor=(0, 0, 0, 0.8), linewidth=0.1, facecolor=(0, 0, 0, 0)),
            Patch(
                edgecolor=(0, 0, 0, 0.8),
                linewidth=0.1,
                facecolor=(0, 0, 0, 0),
                hatch="+++",
            ),
        ],
        labels=["Train", "Test"],
        loc=2,
    )

    cx.add_basemap(ax, source=cx.providers.CartoDB.PositronNoLabels, crs=4326, zoom=11)
    ax.set_axis_off()

    plt.show()

No description has been provided for this image

Loading raw, full data

In [16]:

Copied!

ds = airbnb_multicity.load(version="all")
ds.keys()
ds = airbnb_multicity.load(version="all")
ds.keys()

Out[16]:

dict_keys(['train'])

In [17]:

Copied!

type(airbnb_multicity.train_gdf), type(airbnb_multicity.test_gdf)
type(airbnb_multicity.train_gdf), type(airbnb_multicity.test_gdf)

Out[17]:

(geopandas.geodataframe.GeoDataFrame, NoneType)

In [18]:

Copied!

ds["train"].head()
ds["train"].head()

Out[18]:

	id	name	host_id	host_name	neighbourhood	room_type	price	minimum_nights	number_of_reviews	last_review	reviews_per_month	calculated_host_listings_count	availability_365	number_of_reviews_ltm	city	date	geometry
0	23726706	Private room 20 minutes from Amsterdam + Break...	122619127	Patricia	IJburg - Zeeburgereiland	Private room	88.0	2	78	2022-05-29	1.53	1	66	11	amsterdam	2022-06-05	POINT (4.97879 52.34916)
1	35815036	Vrijstaand vakantiehuis, privé tuin aan het water	269425139	Lydia	Noord-Oost	Entire home/apt	105.0	3	95	2022-06-02	2.65	1	243	36	amsterdam	2022-06-05	POINT (4.95689 52.42419)
2	31553121	Quiet Guesthouse near Amsterdam	76806621	Ralf	Noord-West	Entire home/apt	152.0	2	82	2022-05-29	2.02	1	3	26	amsterdam	2022-06-05	POINT (4.91821 52.43237)
3	34745823	Apartment ' Landzicht', nearby Amsterdam	238083700	Daisy	Gaasperdam - Driemond	Entire home/apt	87.0	2	39	2022-04-17	1.08	3	290	4	amsterdam	2022-06-05	POINT (5.01231 52.2962)
4	44586947	Weesp, 2 kamers vlakbij Amsterdam	360838688	Aranka	Gaasperdam - Driemond	Private room	160.0	2	15	2022-05-29	0.68	1	152	12	amsterdam	2022-06-05	POINT (5.0303 52.31475)

Create your own train-test split -> Spatial splitting with bucket stratification

In [19]:

Copied!

train, test = airbnb_multicity.train_test_split(
    target_column="price", test_size=0.2, resolution=8, n_bins=10, random_state=42
)
train, test = airbnb_multicity.train_test_split(
    target_column="price", test_size=0.2, resolution=8, n_bins=10, random_state=42
)

Summary of the split:

  Train: 33190 H3 cells (2243473 points)
  Test: 26285 H3 cells (560865 points)

  Expected ratios: {'train': 0.8, 'validation': 0, 'test': 0.2}
  Actual ratios: {'train': 0.8, 'test': 0.2}
  Actual ratios difference: {'train': 0.0, 'test': 0.0}
   bucket  train_ratio  test_ratio  train_ratio_difference  \
0       0      0.80000     0.20000                 0.00000   
1       1      0.80000     0.20000                 0.00000   
2       2      0.80000     0.20000                 0.00000   
3       3      0.80000     0.20000                 0.00000   
4       4      0.80000     0.20000                 0.00000   
5       5      0.80000     0.20000                 0.00000   
6       6      0.80000     0.20000                 0.00000   
7       7      0.80000     0.20000                 0.00000   
8       8      0.80000     0.20000                 0.00000   
9       9      0.80001     0.19999                -0.00001   

   test_ratio_difference  train_points  test_points  
0                0.00000        226300        56575  
1                0.00000        231889        57972  
2                0.00000        218161        54541  
3                0.00000        243071        60766  
4                0.00000        206847        51712  
5                0.00000        222598        55650  
6                0.00000        235277        58819  
7                0.00000        212393        53099  
8                0.00000        222589        55646  
9                0.00001        224348        56085

Created new train_gdf and test_gdf. Train len: 2243473,test len: 560865

In [20]:

Copied!

type(airbnb_multicity.train_gdf), type(airbnb_multicity.test_gdf)
type(airbnb_multicity.train_gdf), type(airbnb_multicity.test_gdf)

Out[20]:

(geopandas.geodataframe.GeoDataFrame, geopandas.geodataframe.GeoDataFrame)

In [21]:

Copied!

airbnb_multicity.resolution
airbnb_multicity.resolution

Out[21]:

In [22]:

Copied!

train.head()
train.head()

Out[22]:

	id	name	host_id	host_name	neighbourhood	room_type	price	minimum_nights	number_of_reviews	last_review	reviews_per_month	calculated_host_listings_count	availability_365	number_of_reviews_ltm	city	date	geometry
0	23726706	Private room 20 minutes from Amsterdam + Break...	122619127	Patricia	IJburg - Zeeburgereiland	Private room	88.0	2	78	2022-05-29	1.53	1	66	11	amsterdam	2022-06-05	POINT (4.97879 52.34916)
5	15801253	Studio with own bathroom & kitchen at East A'dam	21813940	Nan	Watergraafsmeer	Private room	90.0	2	46	2022-05-29	0.89	1	164	7	amsterdam	2022-06-05	POINT (4.96413 52.34507)
11	44174770	quiet Apartment at a Lake	114132046	Sven	Watergraafsmeer	Private room	59.0	30	15	2022-03-18	0.66	3	245	10	amsterdam	2022-06-05	POINT (4.95199 52.33715)
12	50888392	Gezellig appartement aan de rand van centrum w...	119648218	Margareth	Gaasperdam - Driemond	Entire home/apt	150.0	2	18	2022-04-28	1.73	1	153	18	amsterdam	2022-06-05	POINT (5.03833 52.30917)
14	48864190	Amsterdam_Farmyard	85831122	Linda & Arjan	IJburg - Zeeburgereiland	Entire home/apt	350.0	1	11	2022-04-04	0.83	3	300	9	amsterdam	2022-06-05	POINT (4.99961 52.33851)

In [23]:

Copied!

test.head()
test.head()

Out[23]:

	id	name	host_id	host_name	neighbourhood	room_type	price	minimum_nights	number_of_reviews	last_review	reviews_per_month	calculated_host_listings_count	availability_365	number_of_reviews_ltm	city	date	geometry
1	35815036	Vrijstaand vakantiehuis, privé tuin aan het water	269425139	Lydia	Noord-Oost	Entire home/apt	105.0	3	95	2022-06-02	2.65	1	243	36	amsterdam	2022-06-05	POINT (4.95689 52.42419)
2	31553121	Quiet Guesthouse near Amsterdam	76806621	Ralf	Noord-West	Entire home/apt	152.0	2	82	2022-05-29	2.02	1	3	26	amsterdam	2022-06-05	POINT (4.91821 52.43237)
3	34745823	Apartment ' Landzicht', nearby Amsterdam	238083700	Daisy	Gaasperdam - Driemond	Entire home/apt	87.0	2	39	2022-04-17	1.08	3	290	4	amsterdam	2022-06-05	POINT (5.01231 52.2962)
4	44586947	Weesp, 2 kamers vlakbij Amsterdam	360838688	Aranka	Gaasperdam - Driemond	Private room	160.0	2	15	2022-05-29	0.68	1	152	12	amsterdam	2022-06-05	POINT (5.0303 52.31475)
6	19572024	Coachhouse, in nature only 5 km from Amsterdam	81955946	Amber	Watergraafsmeer	Entire home/apt	279.0	3	126	2022-05-29	2.13	2	298	23	amsterdam	2022-06-05	POINT (4.90833 52.30739)